English. In this article, we describe our work on porting Open Multilingual Wordnet resources into the OntoLex-Lemon model, in order to establish an interlinking with corresponding morphological resources, such as the MMorph resource set. For this purpose, the morphological resources were also ported onto OntoLexLemon. We show how the “lemmas” contained in the Wordnet resources can be enriched with morphological features using the lexical representation and linking features of OntoLex-Lemon, which support, among others, the formulation of restrictions in the usage of such expressions. Our work will result in an improved lexical resource combining Wordnet senses and full morphological descriptions in a single ontological framework, as specified in the OntoLex-Lemon model.
WordNets are well-established lexical resources
with a wide range of applications. For more than
twenty years they have been elaborately set up
and maintained by hand, especially the original
Princeton WordNet of English (PWN)
Copyright © 2019 for this paper by its authors. Use
permitted under Creative Commons License Attribution 4.0
International (CC BY 4.0).
shared synset IDs
The work described in the present article is an
extension of previous experiments done with
English
Our current work deals primarily with the
morphological enrichment of OMW resources for
Italian, i.e. “ItalWordNet”.2 The first
morphological resource we took into consideration for this
purpose is an updated version of the MMorph
morphological analyser
As a representation mean we chose
OntoLexLemon
OntoLex-Lemon is a further development of
the “Lexicon Model for Ontologies” (lemon)
2See
3See also https://www.w3.org/2016/05/ ontolex/ for more details.
4See https://globalwordnet.github.io/ schemas/##rdf.
5RDF stands for “Resource Description Framework”. See
mapped onto the former lemon model
In the next sections, we give some background information on OMW and MMorph. We continue with a section on OntoLex-Lemon, followed by sections that describe how OntoLex-Lemon supports the linking of lemmas in the OMW resources to full morphological descriptions. Doing so, morphological descriptions can be associated with the conceptual entries of WordNet. 2
OMW is an initiative that brings together Wordnets in different languages, linking them to the original Princeton WordNet (PWN). As stated on the web page of OMW, those Wordnets were of different quality, and some of those were in fact extracted from different types of language resources. We are dealing with three OMW WordNet resources.7 OMW provided for an harmonization of such resources, and published them in a uniform format, which is displayed just below, showing here a few examples from the Italian resource: 08388207 n i t a : lemma n o b i l t a` 08388207 n i t a : lemma a r i s t o c r a z i a 08388207 n i t a : lemma p a t r i z i a t o 08388207 n i t a : d e f 0
l ' i n s i e m e d e g l i a r i s t o c r a t i c i 08388207 n i t a : d e f 1
l ' i n s i e m e d e i n o b i l i https://www.w3.org/RDF/ for more details.
6OntoLex-Lemon is indeed representing an ontology of lexical elements.
7French, Spanish and Italian, with a focus on the latter.
See http://compling.hss.ntu.edu.sg/omw/ for
downloading the resources. For more details see also
As the reader can see in the two examples above, OMW resources deliver information on the synset number, together with the part-of-speech of the associated lemma. In some cases, definitions (marked with ita : def) are provided, as well as examples (marked with ita : exe).
This format is used for all languages of the OMW corpus. This eases its mapping to a formal representation supporting the interoperability and interlinking of language resources, such as the OntoLex-Lemon model (see Section 4). 3
MMorph was originally developed by ISSCO at the University of Geneva in the past MULTEXT project8. For our purposes, we used the extended MMorph version developed at DFKI LT Lab (MMorph3). This version includes huge lexical resources for English, French, German, Italian and Spanish. Very generally, the tool relates a word to a morphosyntactic description (MSD) containing free-definable attribute and values. The MMorph lexicon which is used to realize such MSD consists of a set of lexical entries and structural rules.9 For example, the following rule creates in Italian a noun plural concatenating the noun stem and the gender-specific suffixes: Listing 1: Rule for noun plural generation in Italian. Note how the rule ensures that the gender doesn’t change in the plural form.
N . ms : ” o ” N S u f f i x [ num= s i n g gen =masc t y p e =oa ] N . mp : ” i ” N S u f f i x [ num= p l u r gen =masc t y p e =oa ] N . f s : ” a ” N S u f f i x [ num= s i n g gen =fem t y p e =oa ] N . f p : ” e ” N S u f f i x [ num= p l u r gen =fem t y p e =oa ] FlexN : Noun [ gen =$1 num=$2 form = s u r f ] < Noun [ gen =$1 num= s i n g
form = stem t y p e =$T ] N ASfix [ gen =$1 num=$2
t y p e =$T ] This rule will apply only to the lexical entries (feminine and/or masculine nouns) matching the defined features, e.g.
Noun [ gen =masc num= s i n g form = stem
t y p e =oa ] ” p a t r i z i a t ” = ” p a t r i z i a t o ” ” s u o l ” = ” s u o l o ” 8See https://www.issco.unige.ch/en/ research/projects/MULTEXT.html for more details on the resulting MMorph2:3:4 version.
9See
The MMorph lexica can be dumped to full form lists for the usage in further programs, as can be seen in the following examples: ” n o b i l t a` ” = ” n o b i l t a` ”
Noun [ gen =fem num= s i n g j p l u r ] ” s u o l i ” = ” s u o l o ”
Noun [ gen =masc num= p l u r ] ” s u o l o ” = ” s u o l o ”
Noun [ gen =masc num= s i n g ] The entries above are completed by labelled features for gender and number, but the user can freely define further features, if needed (e.g. clitics for verbal entries or rection of prepositions). Multiple values of a feature are expressed by “j”.
Because of their well-structured form, the dumped Mmorph lexica are ideally suited for the mapping into the OntoLex-Lemon format. 4
The OntoLex-Lemon model was originally developed with the aim to provide a rich linguistic grounding for ontologies, meaning that the natural language expressions used in the description of ontology elements are equipped with an extensive linguistic description.10 This rich linguistic grounding includes the representation of morphological and syntactical properties of a lexical entry as well as the syntax-semantics interface, i.e. the meaning of these lexical entries with respect to an ontology or to specialized vocabularies.
The main organizing unit for those linguistic descriptions is the lexical entry, which enables the representation of morphological patterns for each entry (a MWE, a word or an affix). The connection of a lexical entry to an ontological entity is marked mainly by the denotes property or is mediated by the LexicalSense or the LexicalConcept properties, as represented in Figure 1, which displays the core module of the model.
OntoLex-Lemon is based on and extends the
lemon model
11SKOS stands for “Simple Knowledge Organization Sysin Figure 1, lexical entries can be linked, via the ontolex : evokes property, to such SKOS concepts, which can represent WordNet synsets. This structure parallels the relation between lexical entries and ontological resources, which is implemented either directly by the ontolex : reference property or mediated by the instances of the ontolex : LexicalSense class.12 The ontolex : LexicalConcept class seems to be most appropriate to model the “sets of cognitive synonyms (synsets)”13 described by Princeton WordNet (PWN), while the ontolex : LexicalSense class is meant to represent the bridge between lexical and ontological entities. 5
As mentioned above, the format generated by the OMW initiative is very convenient to map diftem”. SKOS provides “a model for expressing the basic structure and content of concept schemes such as thesauri, classification schemes, subject heading lists, taxonomies, folksonomies, and other similar types of controlled vocabulary” (https://www.w3.org/TR/skos-primer/) 12Quoting from Section 3.6 “Lexical Concept” https: //www.w3.org/2016/05/ontolex/: “We [...] capture the fact that a certain lexical entry can be used to denote a certain ontological predicate. We capture this by saying that the lexical entry denotes the class or ontology element in question. However, sometimes we would like to express the fact that a certain lexical entry evokes a certain mental concept rather than that it refers to a class with a formal interpretation in some model. Thus, in lemon we introduce the class Lexical Concept that represents a mental abstraction, concept or unit of thought that can be lexicalized by a given collection of senses. A lexical concept is thus a subclass of skos:Concept.”
13Quoted from https://wordnet.princeton. edu/. ferent information onto more complex representation frameworks. To transform the OWN data onto the OntoLex-Lemon representation, a Python script was used. A design decision was to extract only the synset information and to encode the synsets as instances of the LexicalConcept class of OntoLex-Lemon. As some OWM lemmas are present in the MMorph resources, we just link the synsets to those lemmas, which are encoded as instances of the OntoLex-Lemon LexicalEntry class (see next section). We will need to create new instances of the OntoLexLemon LexicalEntry class for the OWM lemmas not present in the MMorph resources.
We have now 15553 such LexicalConcept instances for Italian. This is due to the fact that we consider only the subset of ItalWordNet that has been curated by OMW. We also noted that we have less instances of the LexicalConcept as lines for each synset in the original files, as the synset indices are represented by a unique URI in OntoLexLemon.
In Listing 2 we show examples of the OntoLexLemon encoding of two synsets for Spanish.14 The lemmas associated with these synsets are “cura”. In Section 7, we explain how the synsets are linked to the lemmas, which are differentiated in the OntoLex-Lemon representation, but not in the original OMW file.
Listing 2: The OntoLex-Lemon representation of two Spanish synsets : s y n s e t s p a w n 13491616 n r d f : t y p e o n t o l e x : L e x i c a l C o n c e p t ; s k o s : inScheme : s p a w n e t . : s y n s e t s p a w n 10470779 n r d f : t y p e o n t o l e x : L e x i c a l C o n c e p t ; s k o s : inScheme : s p a w n e t . 6
To transform the MMorph data into OntoLexLemon we used a Python script including the rdflib module15, which supports the generation of RDF-graphs in rdf : xml, turtle, or other relevant formats. In Listing 3, we show examples of the resulting data for the lemma “viola” in Italian.
14For the representation of OntoLex-Lemon data, we chose the turtle syntax serialization. More on the turtle syntax: https://www.w3.org/TR/turtle/.
15See https://github.com/RDFLib/rdflib for more details.
Listing 3: The OntoLex-Lemon entry for viola : l e x v i o l a f e m a o n t o l e x : L e x i c a l E n t r y ; l e x i n f o : p a r t O f S p e e c h l e x i n f o : noun ; o n t o l e x : c a n o n i c a l F o r m : f o r m v i o l a f ; o n t o l e x : o t h e r F o r m : f o r m v i o l a f p l . : l e x v i o l a m a s c a o n t o l e x : L e x i c a l E n t r y ; l e x i n f o : p a r t O f S p e e c h l e x i n f o : noun ; o n t o l e x : c a n o n i c a l F o r m : f o r m v i o l a m ; : f o r m v i o l a f a o n t o l e x : Form ; l e x i n f o : g e n d e r l e x i n f o : f e m i n i n e ; l e x i n f o : number l e x i n f o : s i n g u l a r ; o n t o l e x : w r i t t e n R e p ” v i o l a ” @it . : f o r m v i o l a f p l a o n t o l e x : Form ; l e x i n f o : g e n d e r l e x i n f o : f e m i n i n e ; l e x i n f o : number l e x i n f o : p l u r a l ; o n t o l e x : w r i t t e n R e p ” v i o l e ” @it . : f o r m v i o l a m a o n t o l e x : Form ; l e x i n f o : g e n d e r l e x i n f o : m a s c u l i n e ; l e x i n f o : number l e x i n f o : p l u r a l ,
l e x i n f o : s i n g u l a r ; o n t o l e x : w r i t t e n R e p ” v i o l a ” @it .
As the reader can observe, we have two lexical entries for the entry “viola”, as this is requested by the OntoLex-Lemon guidelines, following which a word with different grammatical genders should have one lexical entry per gender. “Viola” in feminine is the music instrument, while in masculine it means “violet”. This is in fact an important feature for linking synsets to lemmas having distinct genders, as we will exemplify in Section 7.
The transformation of nominal entries from MMorph to the OntoLex-Lemon format resulted in 21085 instances of the class LexicalEntry for Italian. We still need to consider the lemmas of the OMW resources that are not in MMorph. This is concerning mostly multiword entries in OMW.
We will also investigate the use of other lexical resources, but the current use of the MMorph was motivated by the fact that we could have access to the different languages available in one and the same format, facilitating thus the uniform mapping into OntoLex-Lemon. 7
We see the use of OntoLex-Lemon for
representing WordNets not only as a chance to port
information from one format to another (including the
possibility to publish WordNets in the Linguistic
Linked Opend Data cloud16), but also as an
opportunity to extend the coverage of WordNet
descrip16See http://linguistic-lod.org/
llod-cloud and
OntoLex-Lemon supports this linking in a straightforward manner. As can be seen in Figure 1, there is a property putting a LexicalConcept in relation to a LexicalEntry, i.e. the property evokes and its reverse isEvokedBy. Therefore we just need to add this property to both the OntoLex-Lemon representations of a synset and its corresponding entry. In Listing 4 we show such a case, taking again the word “cura” as an example.
Listing 4: Interlinking a synset and an entry for cura : s y n s e t s p a w n 13491616 n r d f : t y p e o n t o l e x : L e x i c a l C o n c e p t ; s k o s : inScheme : s p a w n e t ; o n t o l e x : e v o k e s : l e x c u r a 1 . : l e x c u r a 1 a o n t o l e x : L e x i c a l E n t r y ; l e x i n f o : g e n d e r l e x i n f o : fem ; l e x i n f o : p a r t O f S p e e c h l e x i n f o : noun ; o n t o l e x : c a n o n i c a l F o r m : f o r m c u r a ; o n t o l e x : o t h e r F o r m : f o r m c u r a p l u r a l ; o n t o l e x : i s E v o k a t e d B y
: s y n s e t s p a w n 1349161 n . : s y n s e t s p a w n 10470779 n r d f : t y p e o n t o l e x : L e x i c a l C o n c e p t ; s k o s : inScheme : s p a w n e t ; o n t o l e x : e v o k e s : l e x c u r a 2 . : l e x c u r a 2 a o n t o l e x : L e x i c a l E n t r y ; l e x i n f o : g e n d e r l e x i n f o : mas ; l e x i n f o : p a r t O f S p e e c h l e x i n f o : noun ; o n t o l e x : c a n o n i c a l F o r m : f o r m c u r a ; o n t o l e x : o t h e r F o r m : f o r m c u r a p l u r a l ; o n t o l e x : i s E v o k a t e d B y
: s y n s e t s p a w n 10470779 n .
Just adding the properties evokes and its reverse isEvokedBy to the corresponding elements in the generated OntoLex-Lemons data sets is providing for this morphological enrichment of the original Wordnets. Once the original (different types of) resources have been mapped onto the OntoLexLemon model, it is very easy to interlink or even to merge them into a richer representation. An extension of this work consists in describing restrictions on the usage of certain Wordnet concepts, as for example in the Italian case of the noun “bene” versus its plural form “beni”, or English “silk” versus the plural form “silks”, which are associated with different and sometimes not shareable meanings.17 We are making use for this of a strategy described in an extension to the core module of Ontolex-Lemon, called “Lexicog”,18 which foresees the description of instances of a class named FormRestriction, so that it is possible to state that a meaning is available only with the use of a specific form, like singular or plural. 8
We described our work on porting Open Multilingual Wordnet resources into the OntoLex-Lemon model, in order to establish an interlinking with corresponding morphological resources, such as the MMorph resource set. For this purpose, the morphological resources were also ported onto OntoLex-Lemon. As a result we noticed that this model can be easily used for bridging the WordNet type of lexical resources to a full description of lexical entries, which coult possibly lead to an extension of the coverage of WordNets beyond the consideration of lemmas and PoS information.
We documented our interlinking work with the example of the full morphological representation of Italian words, putting them in relation with the corresponding OMW data sets. We also started to investigate the description of usage restrictions, which allows us to state formally that certain Wordnet concepts should be used only in the singular or in the plural form.
As a final goal of our work, we see the
interlinked or merged resources in the Linguistic
Linked Open Data (LLOD) cloud. We will
investigate how our work can be combined with
resources present in the LLOD, especially with the
BabelNet framework, which is already
integrating a huge number of lexical resources, including
Princeton WordNet, and encyclopedic data sets
17The reader can see the different meanings associated to those plural words while querying for those in the user interface of PWN: http://wordnetweb.princeton. edu/perl/webwn.
18The current state of this “Lexicography” module is available at https://www.w3.org/community/ ontolex/wiki/Lexicography. The presented work has been supported in part by the H2020 project “Preˆt-a`-LLOD” with Grant Agreement number 825182. Contributions by Thierry Declerck have been supported additionally in part and by the H2020 project “ELEXIS” with Grant Agreement number 731015.